US nursing homes (NH), with their 1.6 million residents, are important sites for the delivery of mental health services as well as for research on the care of the larger group of older, chronically ill, and disabled individuals. With their multiple chronic illnesses and complex medication regimens, NH residents are at high risk for morbidity related to both depression and the side effects of antidepressants (AD). Although treatment is efficacious, even in NH residents, there are concerns that AD, including Serotonin Reuptake Inhibitors (SSRI's), can lead to significant adverse events including falls and fractures. This vulnerability must be viewed in the context of ongoing patterns of care. Nationally, more than 1/3 of NH residents are taking AD. Based on estimates from our research, approximately 14% of cognitively intact or mild-moderately impaired long term care residents have untreated depression, 20% appear to be partially treated, and 26% are taking AD but do not have current depressive symptoms. Most nursing home residents who are depressed are suffering from an initial episode of late life depression, and, therefore, when they go into remission, they do not meet guideline- based recommendations for maintenance treatment. Thus, for substantial numbers of nursing home residents, there are questions about the balance between the benefits of long term treatment with AD's (to prevent recurrences of depression) versus the risks (related to adverse effects including falls, injuries, and fractures). Accordingly, we propose to conduct a partially randomized patient (and provider) preference study for NH residents taking AD's who have been in remission from (at most) a first episode of depression for a period of 6 months or longer. We will randomize 160 individuals who are ambulatory or semi-ambulatory, cognitively intact or with mild-moderate impairment, from multiple nursing homes to continue AD's or to be withdrawn from them under open-label conditions, and to be followed for a period of one year. To improve generalizability, we will also follow otherwise eligible individuals who refuse randomization because they, their families, or providers have a strong preference for one of the conditions. The study is designed to test the hypotheses that continued AD use prevents recurrences, but that it also increases the risk of falls. We will also estimate relevant effect sizes, and, in more exploratory analyses, evaluate the benefits versus the risks of continued AD use with measures of mood, functioning, behavior, gait/balance, and related parameters.